tor hidden services & SSL EV certificate
I was at the talk this afternoon at the 32c3 and <i am wondering where can get a certificate for a .onion. Any service to suggest? Also where I should see to configure it correctly? - benoit
On 12/29/2015 10:25 AM, Benoit Chesneau wrote:
I was at the talk this afternoon at the 32c3 and <i am wondering where can get a certificate for a .onion. Any service to suggest? Also where I should see to configure it correctly?
- benoit
You don't need one. Hidden services automatically get end-to-end authentication and encryption. Since that is handled by Tor and not by the browser, hidden service addresses use "http" rather than "https", but in this case the connection is nevertheless encrypted. It's technically redundant to add HTTPS. A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else. -- Jesse V
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 I beg to differ. One of the very good points made in the talk was that by tying the "vanilla" DNS name of the website and its .onion address as alternate names, you can offer proof to your users that the .onion URL they entered is indeed the website they were trying to reach. Barring that, you have to trust on good faith that the random string you found on Google is not bringing you to a malicious copy of your destination which performs man-in-the-middle to steal your credentials (and/or rewrites Bitcoin address since apparently that's a thing). As for the original question, I think that you cannot get a DV certificate for the .onion TLD at the moment. I assume that you could go the FaceBook way and try your luck with Verisign or Digicert, but it's probably going to cost you a few hundred of dollars. Since you're at 32c3, you should get in touch with the EFF / Let's Encrypt people to see if they have made plans for this issue. - -- Ivan On 12/29/2015 08:38 PM, Jesse V wrote:
On 12/29/2015 10:25 AM, Benoit Chesneau wrote:
I was at the talk this afternoon at the 32c3 and <i am wondering where can get a certificate for a .onion. Any service to suggest? Also where I should see to configure it correctly?
- benoit
You don't need one. Hidden services automatically get end-to-end authentication and encryption. Since that is handled by Tor and not by the browser, hidden service addresses use "http" rather than "https", but in this case the connection is nevertheless encrypted. It's technically redundant to add HTTPS. A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
_______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
-----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBCAAGBQJWgud3AAoJEEDp8Kj16odUcN0P+QElB/86fiE1Sl5nvN8SrkaB Tw9sMpWh3Eciccyu831l8eljb2zsEcsVaGrpye3gKTdCRDHJIw+6zT2clRwgxlby 1isi7/x2rUr18ar2qE32edisTy4++FUJA+qEZADBQGzAojVFMn+MZIODkVBGFmwG raOycdABpCk6GD1z8Xes0vNzFoFws4sEyWuqWWsAYNbDyokHYWSI41MtgqiSuzJe 3pB+SNwZ5IeP3vOjZbpL/pe7zoqqP2eKjPstT50MhOblBQtvbjQPc439meCHjmcu 1sXCNClODZFuyEOa2dQO6aG9fX/nF6xEYkQViF34otgSwcKB6IcIRsaXD+MLT7Ts wS18ug7ZjaMChq59Ug4BtjSJGrQqbjLmzaYq+Sk8GCVXVxgLAJqfQzDEDi3k3nBE xPdeDYYaD5poGlfevpSi6xdmI5N0jbBv74TI9XuLtDgg6PCI1oUccH+O+Qyfabn/ al81ANcQsTYgTMDb6iemo7HJ7FxN+1fHHy8voKecDG8hzJUUDMh/rXzrt8egbWbj y0Q0wbW7vY+AjJy+YYrWe5YHCxJ9u7VacJ59YCkFPeG7ZXIUqv15cKTxVMXhgeR+ ZhksZgWSQ1bzbThMRP/alrPvSL2N9yhAxXkdQ+ycKzNygXS88M7S0OIKBIAp73zO MY6YnhupRj2N1+9X8dqe =mCnL -----END PGP SIGNATURE-----
On 29 Dec 2015, at 21:05, Ivan Kwiatkowski <justicerage@manalyzer.org> wrote:
As for the original question, I think that you cannot get a DV certificate for the .onion TLD at the moment. I assume that you could go the FaceBook way and try your luck with Verisign or Digicert, but it's probably going to cost you a few hundred of dollars.
Just for info ordering a .onion certificate in digicert is supported: https://blog.digicert.com/ordering-a-onion-certificate-from-digicert/ - benoit
On 29/12/2015 19:38, Jesse V wrote:
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
(I am not commenting on the technical necessity of a cert.) No, I think the point that was made at today's talk (and correct me if I got it wrong) was that if I am the operator of, for example, www.bigclearwebwebsite.com (who, by default of big known to the Internet, I am not worried about the anonymity of my site or those who operate it). I want to create a www.bigclearwebwebsite.onion site (which of course would be more like www.xhsjeflflajdfyeysksldpfiejcc.onion), I can do this by getting a HTTPS cert for my .onion address. The objective of it (from a users point of view) would be the tieing the identity of the *clear web* site and the *.onion site* together to give the user some trust that bigclearwebwebsite.onion is in fact the same as the .com site. (Replace bigclearwebwebsite. with DuckDuckGo, Facebook, etc)
On 12/29/2015 01:16 PM, bernard wrote:
On 29/12/2015 19:38, Jesse V wrote:
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
(I am not commenting on the technical necessity of a cert.)
No, I think the point that was made at today's talk (and correct me if I got it wrong) was that if I am the operator of, for example, www.bigclearwebwebsite.com (who, by default of big known to the Internet, I am not worried about the anonymity of my site or those who operate it).
I want to create a www.bigclearwebwebsite.onion site (which of course would be more like www.xhsjeflflajdfyeysksldpfiejcc.onion), I can do this by getting a HTTPS cert for my .onion address.
The objective of it (from a users point of view) would be the tieing the identity of the *clear web* site and the *.onion site* together to give the user some trust that bigclearwebwebsite.onion is in fact the same as the .com site.
(Replace bigclearwebwebsite. with DuckDuckGo, Facebook, etc)
True. But I don't see that it helps much for onion sites that aren't tied to well-known clearweb sites. Spoofers could also get HTTPS certs. And users couldn't tell them apart. I've been playing with GnuPG-signed pages, with the public key available from multiple independent sources. But of course, it's a bit much to expect users to verify signatures.
On 29/12/2015 20:55, Mirimir wrote:
On 12/29/2015 01:16 PM, bernard wrote:
<snip>
The objective of it (from a users point of view) would be the tieing the identity of the *clear web* site and the *.onion site* together to give the user some trust that bigclearwebwebsite.onion is in fact the same as the .com site.
(Replace bigclearwebwebsite. with DuckDuckGo, Facebook, etc)
True. But I don't see that it helps much for onion sites that aren't tied to well-known clearweb sites.
Fair point. Maybe for users of a .onion site that *itself* needs anonymity, an SSL cert is not helpful to it. But another way of asking: how do I as a user trust that .onion site?
I've been playing with GnuPG-signed pages, with the public key available from multiple independent sources. But of course, it's a bit much to expect users to verify signatures.
Well, given difficulties users have with verification of GPG key strings, that would be difficult. Bernard
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
As indicated in the roger’s lecture, HTTPS is usefull for HS : - browsers handle more securely cookies or other stuff in HTTPS mode, avoiding some possible leaks - because anybody can create an HS and proxify any content, X.509 certs allow users to verify the authenticity of the HS (you are on the official Facebook HS if you have a cert with facebook.com *AND* facebookcorewwwi.onion inside) -- Aeris Individual crypto-terrorist group self-radicalized on the digital Internet https://imirhil.fr/ Protect your privacy, encrypt your communications GPG : EFB74277 ECE4E222 OTR : 5769616D 2D3DAC72 https://café-vie-privée.fr/
On 29 Dec 2015, at 21:18, Aeris <aeris+tor@imirhil.fr> wrote:
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
As indicated in the roger’s lecture, HTTPS is usefull for HS : - browsers handle more securely cookies or other stuff in HTTPS mode, avoiding some possible leaks - because anybody can create an HS and proxify any content, X.509 certs allow users to verify the authenticity of the HS (you are on the official Facebook HS if you have a cert with facebook.com *AND* facebookcorewwwi.onion inside)
Yes I forgot to say it was for a know url already (https://friendpaste.com) on which I would like to add a .onion access so people don't have to go out of the tor network somehow. Having a cert working on that address would help to validate the access. I will check with digicert if this is possible since they are already providing me a cert for this address. - benoit
On 12/29/2015 11:18 AM, Aeris wrote:
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
As indicated in the roger’s lecture, HTTPS is usefull for HS : - browsers handle more securely cookies or other stuff in HTTPS mode, avoiding some possible leaks - because anybody can create an HS and proxify any content, X.509 certs allow users to verify the authenticity of the HS (you are on the official Facebook HS if you have a cert with facebook.com *AND* facebookcorewwwi.onion inside)
I've downloaded the .webm of Roger's lecture but haven't had the time today to listen to it. My point was that HSs already have an authentication mechanism and it's assumed that you can verify the address through some trusted out-of-band method, so in that case you don't need an SSL cert. This can sometimes be superior to trusting the centralized CA model, but I agree that the points you've listed are useful applications as well. -- Jesse V
There should be a way to auth via letsencrypt.org, anonymously. To: tor-relays@lists.torproject.org From: kernelcorn@riseup.net Date: Tue, 29 Dec 2015 12:27:06 -0900 Subject: Re: [tor-relays] tor hidden services & SSL EV certificate On 12/29/2015 11:18 AM, Aeris wrote:
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
As indicated in the roger’s lecture, HTTPS is usefull for HS : - browsers handle more securely cookies or other stuff in HTTPS mode, avoiding some possible leaks - because anybody can create an HS and proxify any content, X.509 certs allow users to verify the authenticity of the HS (you are on the official Facebook HS if you have a cert with facebook.com *AND* facebookcorewwwi.onion inside)
I've downloaded the .webm of Roger's lecture but haven't had the time today to listen to it. My point was that HSs already have an authentication mechanism and it's assumed that you can verify the address through some trusted out-of-band method, so in that case you don't need an SSL cert. This can sometimes be superior to trusting the centralized CA model, but I agree that the points you've listed are useful applications as well. -- Jesse V _______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
On Tue, Dec 29, 2015 at 12:27:06PM -0900, Jesse V wrote:
On 12/29/2015 11:18 AM, Aeris wrote:
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
As indicated in the roger’s lecture, HTTPS is usefull for HS : - browsers handle more securely cookies or other stuff in HTTPS mode, avoiding some possible leaks - because anybody can create an HS and proxify any content, X.509 certs allow users to verify the authenticity of the HS (you are on the official Facebook HS if you have a cert with facebook.com *AND* facebookcorewwwi.onion inside)
I've downloaded the .webm of Roger's lecture but haven't had the time today to listen to it. My point was that HSs already have an authentication mechanism and it's assumed that you can verify the address through some trusted out-of-band method, so in that case you don't need an SSL cert. This can sometimes be superior to trusting the centralized CA model, but I agree that the points you've listed are useful applications as well.
In case it is helpful. Griffin Boyce and I have a paper forthcoming in IEEE Security & Privacy Magazine on this topic. The final editorial changes are not in so it might change a little, but you can find the hopefully-close-to-final version at https://github.com/saint/w2sp-2015/blob/master/SP_SPSI-2015-09-0170.R1_Syver... It covers - How the self-authentication of onionsites that Jesse has been noting and the SSL certs for registered-domain websites that Benoit asked about can complement each other in a variety of ways---and not just for big companies but for individuals, small businesses, local organizations, clubs, sports teams, etc. - The current state of certs for onionsites (EV only), and what the issues are that stand in the way of DV certs and a proposal for resolving them. - How this can all dovetail nicely with Let's Encrypt (an issuance and usage design that binds things together nicely so it is hard to undetectably set up a spoof onionsite of another onionsite of a registered-domain site, etc. and vice versa) once DV certs are allowed. - A description of using GPG that can be done right now while waiting for the world to catch up, and an existing example of a site that does such binding (from a small site operator who found his hosting provider was blocking access from the Tor network). We just cited one such example in the paper, but there are of course others, e.g., https://blog.patternsinthevoid.net/isis.txt aloha, Paul
On 30 Dec 2015, at 13:55, Paul Syverson <paul.syverson@nrl.navy.mil> wrote:
On Tue, Dec 29, 2015 at 12:27:06PM -0900, Jesse V wrote:
On 12/29/2015 11:18 AM, Aeris wrote:
A few hidden services have added an HTTPS cert but I think that's mostly for a publicity stunt than anything else.
As indicated in the roger’s lecture, HTTPS is usefull for HS : - browsers handle more securely cookies or other stuff in HTTPS mode, avoiding some possible leaks - because anybody can create an HS and proxify any content, X.509 certs allow users to verify the authenticity of the HS (you are on the official Facebook HS if you have a cert with facebook.com *AND* facebookcorewwwi.onion inside)
I've downloaded the .webm of Roger's lecture but haven't had the time today to listen to it. My point was that HSs already have an authentication mechanism and it's assumed that you can verify the address through some trusted out-of-band method, so in that case you don't need an SSL cert. This can sometimes be superior to trusting the centralized CA model, but I agree that the points you've listed are useful applications as well.
In case it is helpful. Griffin Boyce and I have a paper forthcoming in IEEE Security & Privacy Magazine on this topic. The final editorial changes are not in so it might change a little, but you can find the hopefully-close-to-final version at https://github.com/saint/w2sp-2015/blob/master/SP_SPSI-2015-09-0170.R1_Syver... <https://github.com/saint/w2sp-2015/blob/master/SP_SPSI-2015-09-0170.R1_Syverson.pdf>
It covers
- How the self-authentication of onionsites that Jesse has been noting and the SSL certs for registered-domain websites that Benoit asked about can complement each other in a variety of ways---and not just for big companies but for individuals, small businesses, local organizations, clubs, sports teams, etc.
- The current state of certs for onionsites (EV only), and what the issues are that stand in the way of DV certs and a proposal for resolving them.
- How this can all dovetail nicely with Let's Encrypt (an issuance and usage design that binds things together nicely so it is hard to undetectably set up a spoof onionsite of another onionsite of a registered-domain site, etc. and vice versa) once DV certs are allowed.
- A description of using GPG that can be done right now while waiting for the world to catch up, and an existing example of a site that does such binding (from a small site operator who found his hosting provider was blocking access from the Tor network). We just cited one such example in the paper, but there are of course others, e.g., https://blog.patternsinthevoid.net/isis.txt <https://blog.patternsinthevoid.net/isis.txt>
aloha, Paul _______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org <mailto:tor-relays@lists.torproject.org> https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays <https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays>
Thanks it's useful :) I am know wondering how i can bruteforce a clear name for my site like facebook but i think it's all good for the rest :) - benoît
On 12/31/2015 05:44 AM, Benoit Chesneau wrote:
Thanks it's useful :) I am know wondering how i can bruteforce a clear name for my site like facebook but i think it's all good for the rest :)
Like this: https://github.com/katmagic/Shallot It doesn't look like it's maintained, so you may want to examine forks, but I've had success with Shallot as it is. -- Jesse V
participants (9)
-
Aeris -
Benoit Chesneau -
benoitc
-
bernard -
Billy Humphreys -
Ivan Kwiatkowski -
Jesse V -
Mirimir -
Paul Syverson